PCB Copper Trace Width for 1A

[made2hack]

Hi All,

According to this PCB Trace calculator, if I want 2A through my circuit, I should have trace widths of approximately 2mm.

My pcb board is the standard 1.6mm thick, 35um (1oz / ft2) type, one sided.

If I want to pass 40V / 1A throughout, is 2A head room enough? And, I suspect that only the path of the high current / high voltage needs to be this thick? Other traces such as to resistors, or ref voltage pins doesn't need to be this thick? Or should I make it all this thick?

Should I have ground plane also as thick as possible? I've also got a SOT89-5 ic with a large ground plane tab?

[Rudane]

I'd need to see a schematic, but any trace that is going to carry that high current needs to be "beefed up" or widened so that it won't burn up on you. If you are sure the maximum current will be 1A, then 2A is fine, but the question you should ask is: "how confident are you that the maximum current is 1A?" The general rule in engineering is if you have very little confidence, use a factor of 4, and if you are very confident use a factor of like 1.5 to 2.

Personally, I would multiple by two, and design it for 2A, but if this was not being mass produced I would leave the soldermask off the trace in question and put a layer of solder on it when I was building the circuit. This layer of solder will increase the maximum current and reduce the resistance of the trace. Essentially you would be increasing the "copper thickness" of 1 oz/ft^2 for that trace only.

Again, only the traces that will carry high current need to be modified.

When you ask about the ground plane being as thick as possible, do you mean the area of it as big as possible? If so, then yes the ground plane should be as large as you can make it without sacrificing performance or price. I put ground everywhere there isn't a trace on my boards (fill).

And always remember, voltage appears across and current flows through.

[T3sl4co1l]

It is what it is.  You select the temp rise and whether it's an internal or external layer trace.

You can put tens of amperes through such a trace, day in day out -- but you'd better keep it cool, or run at a low pulsed duty cycle.  It might fuse usefully fast in the 50-100A range, depending on how long it is (connections to fatter copper and component pads will effectively heatsink the trace).  Or if you require low resistances, well below what the temp rise alone demands, you might need massive traces, or special routing techniques (kelvin leads, differential measurements -- even within the same board, etc.).

A 30 mil (not quite 1mm) trace for 1A, not doing anything special (no concerns over power loss, voltage drop, etc.), is quite reasonable.

Tim

[made2hack]

@Rudane, you can have a look at the schematic in this post: https://www.eevblog.com/forum/beginners/12v-to-40v-boost-using-an-lt1070/

Also the discussion about what components I used.

[made2hack]

What I mean about the gnd plane, yeah the surface, as the thickness won't change. I mean, can I fill all the unused "PCB" with GND? It won't affect the circuit negatively?

[made2hack]

Sorry, that post is TL:DR.

Here's the final schematic:

[tszaboo]

Here you have to consider almost everything high current. Isense not, but everything else should be able to carry that current. 2mm is not that horrible, don't worry. It only gets interesting above 10A.

[T3sl4co1l]

I normally flood with ground pour.  Makes the circuit better.  Ideally, you get 4-layer performance for 2-layer cost.  You do have to stitch the ground planes religiously, otherwise every little gap (a trace running through the pour) makes it not be ground plane.

Be sure to have a ground plane underneath the switching traces, and put the bypass caps very close by.

Note... is that really a 40V supply and a 20V schottky diode?  Rather than schottky (of any voltage), a high speed 200V junction diode would be more typical at that voltage.  If schottky is insisted upon, it should be an 80V device.

Tim

[made2hack]

@Tim,

My initial design was for a 20V diode since I misread the datasheet for the led driver. As per @mariush's help, I chose a Fairchild DIODE, SCHOTTKY, 5A, 60V, DO-201

[made2hack]

@Tim,

This is only a 1 sided board, that's why I wasn't sure whether the "ground plane" such as the one on the SOT89's pad and everything else which goes to ground (and the V-) should be connected together, such that the ground plane "returns" via the V- all the way to the source voltage.

[T3sl4co1l]

Single sided hmm... if you have enough clearance between pads to allow ground fill, it can be almost as good.  Not really, but better than a 2-layer board with thin traces everywhere and no ground plane.  But if you've got a lot of close spaced rows of pads, that isn't usually possible, and you end up with ground looped around areas it can't reach across.  To some extent, you can use jumpers to stitch that, but...

On the upside, if you're buying boards, basically every place offers two sided proto as standard because it's what everyone gets, and you'll often spend more (or more lead time anyway) to get something else, like single sided.

Tim

[Phaedrus]

Here you have to consider almost everything high current. Isense not, but everything else should be able to carry that current. 2mm is not that horrible, don't worry. It only gets interesting above 10A.

We have a custom breakout test PCB designed to carry up to 100A @+12V. It's split into ten channels with 2.5cm wide traces with solder on the backplane, with screw terminal connections for the 14ga wires.

[Rudane]

The best thing to do is make all the traces there the required 2 mm.

[made2hack]

100A @ 12V? Awesome.

Did anything "weld" to the traces?

[made2hack]

@All

I forgot to mention that this is a pcb that I am etching myself. So, just plain copper, no solder mask. Is there a potential for a solder bridge to occur?

[T3sl4co1l]

As in just out of the blue?  No.  Any shorts will be entirely your fault, and will probably be painfully obvious.

Tim

[made2hack]

Ahh, ok. Cool 

[Phaedrus]

100A @ 12V? Awesome.

Did anything "weld" to the traces?

No. I did once accidentally short the back of an SMA connector, I was in a hurry and missed a pair of touching solder blobs. That connector was connected to the +12V via a 2mm trace. When I powered up the 1300W PSU under test the board went "PAP" and there was a puff of smoke a couple seconds later. Fortunately I didn't have the scope connected yet... The trace burned clean through and lifted clean off the board. I had to patch it with a length of tinned copper wire.

[made2hack]

Do I smell Trace Removal Tool???